Steelcase’s CTO says the AI boom will reshape office design
Steve Miller joined office furniture manufacturer Steelcase in 1999 as a software developer, coinciding with the tail end of the dot-com boom times that birthed future tech giants like Amazon and eBay, as well as plenty of flops like Pets.com and Webvan.
But regardless of the ultimate resiliency of those businesses, the dot-com wave led to a proliferation of technology tools across nearly all sectors. Employers embraced open plan office layouts, ripping out private rooms in favor of spaces that promoted greater collaboration. Desks were redesigned to factor in increased usage of laptops and dual-monitor workstations. And now, as AI usage steadily increases among U.S. employees, Miller anticipates that office design will change yet again.
“This AI super cycle is changing the way people are working,” says Miller, chief technology officer at Steelcase, which was acquired by rival HNI in a $2.2 billion transaction that closed in December. The merger is a bet that the larger, combined company can benefit from stricter return-to-office policies.
Miller says the company has forged partnerships with tech giant Microsoft and Logitech, a Swiss maker of peripheral computer products like keyboards, headphones, and mice, to better understand how AI is changing work and what workspaces employees will need to support their new responsibilities. Steelcase says that furniture, acoustics, camera positioning, and lightning will all need to be reconsidered
A few focus areas that Steelcase has settled on include acoustically private spaces that will make it easier for employees to use AI to record their calls, team-focused spaces that will allow humans and AI systems on a screen to collaborate on work projects, and rejuvenation spaces meant to give workers a break.
On the latter, Steelcase cites Quantum Workplace data that shows frequent AI users report higher levels of burnout (45%) compared to those that use the technology infrequently (38%). This week, Harvard Business Review shared a preview of some in-progress research on the topic, finding that as AI tools broaden the scope of work and allow employees to work at a faster pace, these productivity gains frequently come with fatigue and weaker decision-making.
When vetting potential AI solutions that can be utilized within Steelcase’s own four walls, Miller says a cross-functional, “community practice” group of around 600 employees provides oversight into which AI use cases should be explored and implemented. Input is shared from workers all across the company, including finance, engineering, sales, and operations. A data governance council was also set up to monitor and enforce Steelcase’s guardrails.
Steelcase has deployed some productivity tools across the organization, including Microsoft Copilot, though Miller says he prioritizes larger impact use cases. AI usage metrics aren’t enough to convince him that the technology is worth investing in. “Adoption isn’t transformation,” says Miller. “You can have a lot of activity that isn’t actually doing anything.”
One bigger bet that Steelcase launched more than a year ago is Casey AI Assistant. Developed with Microsoft, the tool trains AI models on Steelcase’s research, furniture spec guides, and product data to speed up the process of creating custom configurations. Designers are given a lot of freedom to create furniture and office space products for each unique client. Infusing AI, Miller says, is “making it possible for designers to navigate that incredible amount of choice and find solutions that match the customer that we’re trying to work for.”
Steelcase’s data shows Casey AI Assistant had a 72% repeat user rate. Casey has 4,700 users who have conducted 58,000 total conversations, averaging about 250 conversations per day.
Miller isn’t yet ready to publicly share any major details about Steelcase’s agentic AI efforts. “We put a lot of energy at the moment into creating our program for vetting and launching agentic AI solutions so that they respect data security and data governance,” he says. “That’s what we’ve been working pretty closely with Microsoft on getting that properly established.”
What’s held back adoption of agentic thus far are Miller’s concerns that agents could override the controls that IT would build into the system. He also wants to be sure that multi-agent systems, which string together multiple AI agents to share information and make decisions together, are also sharing the right data and producing safe and reliable outputs.
“That’s what we want to make sure that we’re getting right,” says Miller.
John Kell
Send thoughts or suggestions to CIO Intelligence here.
This story was originally featured on Fortune.com